Low Particle Swap Rates GEMC
- piskuliche
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 30
- Thank you received: 1
7 years 4 months ago #263
by piskuliche
Hello,
I have been running into a small problem with some of my Gibbs Ensemble Monte Carlo calculations and I was hoping that you might have suggestions for how I might better run these calculations.
In my GEMC calculations, I am seeing for certain systems (using Ewald summation) that volume changes are being preferred when compared with swaps (volume changes having an acceptance rate of about 8-10%, and particle swaps having acceptance rates closer to 0.1-0.5%).
The problem with this - is that my box volumes are fluctuating quite a bit and box compositions are not changing very much. These large volume fluctuations are causing the system to crash (when fluctuations cause the box length to become less than twice the cutoff). Is there a way of avoiding this problem without just increasing the system size (which significantly slows down these calculations)? Thanks for your help in advance.
I have been running into a small problem with some of my Gibbs Ensemble Monte Carlo calculations and I was hoping that you might have suggestions for how I might better run these calculations.
In my GEMC calculations, I am seeing for certain systems (using Ewald summation) that volume changes are being preferred when compared with swaps (volume changes having an acceptance rate of about 8-10%, and particle swaps having acceptance rates closer to 0.1-0.5%).
The problem with this - is that my box volumes are fluctuating quite a bit and box compositions are not changing very much. These large volume fluctuations are causing the system to crash (when fluctuations cause the box length to become less than twice the cutoff). Is there a way of avoiding this problem without just increasing the system size (which significantly slows down these calculations)? Thanks for your help in advance.
Please Log in to join the conversation.
- ryangmullen
- Offline
- Administrator
Less
More
- Posts: 124
- Karma: 4
- Thank you received: 24
7 years 4 months ago #264
by ryangmullen
Hi Zeke,
During equilibration, you can avoid the crashes you describe by setting the cutoff to always be equal to half the box length. You can turn on that option in the # VDW_Style section by placing "true" as the fourth argument. For example,
# VDW_Style
lj cut_tail 60.0 true
lj cut_tail 14.0
will always use the half box length as the cutoff for box 1, but will use a fixed 14.0 Angstrom cutoff for box 2. Since this option violates detailed balance, it should only be used during equilibration.
As for your relative acceptance rates, if you want to have 1 accepted volume change per sweep (at 8-10% acceptance), you can make 10-12 attempts per sweep. If you want 100 accepted molecule swaps per sweep (at 0.1-0.5% acceptance), then you'll need to make 20,000-100,000 attempts.
Best of luck!
Ryan
During equilibration, you can avoid the crashes you describe by setting the cutoff to always be equal to half the box length. You can turn on that option in the # VDW_Style section by placing "true" as the fourth argument. For example,
# VDW_Style
lj cut_tail 60.0 true
lj cut_tail 14.0
will always use the half box length as the cutoff for box 1, but will use a fixed 14.0 Angstrom cutoff for box 2. Since this option violates detailed balance, it should only be used during equilibration.
As for your relative acceptance rates, if you want to have 1 accepted volume change per sweep (at 8-10% acceptance), you can make 10-12 attempts per sweep. If you want 100 accepted molecule swaps per sweep (at 0.1-0.5% acceptance), then you'll need to make 20,000-100,000 attempts.
Best of luck!
Ryan
Please Log in to join the conversation.
- piskuliche
- Topic Author
- Offline
- Junior Member
Less
More
- Posts: 30
- Thank you received: 1
Time to create page: 0.105 seconds